12
$\begingroup$

I have stumbled upon a construction that has probably been noticed before, and I wonder if anyone can point me to a reference.

Suppose that $K$ is a simplicial complex. Let $P(K)$ be the free abelian group on the set of simplices of $K$, all dimensions together. View it as a Grothendieck group of subcomplexes, so that for example the boundary $\partial \sigma$ of a simplex $\sigma$ appears in it as an alternating sum of the proper faces of $\sigma$. The open simplices $int (\sigma) =\sigma-\partial \sigma$ form another basis. Do not think about orientations.

Let $I$ be the involution of $P(K)$ given by $\sigma\mapsto (-1)^m\ int\ (\sigma)$ where $m$ is the dimension of $\sigma$.

It appears that the homology of the $2$-periodic chain complex given by $1-I:P(K)\to P(K)$ and $1+I:P(K)\to P(K)$ is isomorphic to the mod $2$ homology of $K$ made $2$-periodic.

EDIT: In case the slightly vague reference to a "Grothendieck group" is confusing, let me be more explicit: $P(K)$ has a basis consisting of the simplices of the complex. If $\sigma$ is a $p$-simplex of $K$, then the map $I$ takes the basis element $\sigma$ to the sum of $2^{p+1}-1$ terms $(-1)^{dim\ \tau}\tau$, one term for each simplex $\tau$ that is contained in $\sigma$, including $\sigma$ itself. Note that I never mentioned an orientation or an ordering of vertices.

$\endgroup$
  • $\begingroup$ Could you clarify the definition of $P(K)$ some more? I'm not sure what you mean when you say that the boundary of $\sigma$ appears "in it" as an alternating sum. In short, how is $P(K)$ different from the direct sum of all the simplicial chain groups $C_m(K)$? $\endgroup$ – Vidit Nanda Mar 21 '14 at 18:57
  • $\begingroup$ Your $I$ is $K$ of Verdier duality. Thus, the homology of $1-I$ is about self-dual sheaves. Cross-referencing $L$-theory and 2-periodic 2-torsion suggests the Rothenberg exact sequence for change of decoration. $\endgroup$ – Ben Wieland Mar 31 '14 at 18:39
  • $\begingroup$ Oh, I didn't know what Verdier duality is. I suppose I'm looking at cosheaves rather than sheaves but it's otherwise the same. This formalism was serving me nicely even though I wasn't using any derived functors as far as I can tell. $\endgroup$ – Tom Goodwillie Apr 1 '14 at 22:49
  • $\begingroup$ There is an obvious contravariant equivalence between sheaves and cosheaves on a fixed stratification with values in perfect complexes. Verdier duality is best thought of as a weird covariant equivalence, $F\mapsto\{U\mapsto H_c^*(U;F)\}$. How did you get $I$ (or $1-I$) without derived functors? Maybe you should have had them. $\endgroup$ – Ben Wieland Apr 2 '14 at 22:11
  • $\begingroup$ Ben, it's just what I said. And then you can take a direct limit over subdivision to get something that depends on a PL space instead of a simplicial complex. $\endgroup$ – Tom Goodwillie Apr 3 '14 at 23:26

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.